[okl4-developer] Starting with graphic output

[Geoffrey Lee]

On Thu, Apr 24, 2008 at 04:25:03PM +0000, Remy Gottschalk wrote:
> Hi Nelson

Hi

> > 
> > > While using ia32_pc99_vga i encountered another problem that may be
> > > related to my misunderstanding. When i try to access the frame  
> > > buffer. I
> > > can open and mmap the device successfully but when I try to write in  
> > > it
> > > I get a roottask pagefault :
> > >
> > > roottask read pagefault at 00000018, ip=00206d7f - deny
> > > --- "KD# roottask pagefault" ---
> > > --------------------------------- (eip=f0106f79, esp=f0110300) ---
> > 
> > Hm.  Short of getting your binary and performing an object dump on the  
> > faulting IP, it would be a bit hard to pinpoint where the problem may  
> > be.  Let us know if there's anything we can help in that regard.
> > 
> 
> Got a little hint when switching to okl4 v2.1. Now the problem is that
> mmaping the framebuffer fail returning a "permission denied" error.

When you mmap device objects in Linux user space on OKL4, the mapping code
in Iguana checks whether there is an actual object backing that physical
adddress.  For actual physical memory that's easy.  For devices
you need to add the appropriate memory range at build time
in the machine configuration.

As a quick fix you may be able to add a quick hack to eas_map() to 
add backdoor range check for your framebuffer physical range.

By the way, the roottask pagefault is due to an invalid memory
dereference at 0x18.  This is something we know about and we 
are already tracking this as a bug in our bug database.

	-gl

> 
> 
> -- 
> Remy Gottschalk - rgottschalk at linagora.com
> Ingénieur informatique embarquée
> Groupe LINAGORA - http://www.linagora.com
> Tél.: +33(0)1 58 18 68 28 - Fax : +33(0)1 58 18 68 29
> 
> 
> _______________________________________________
> Developer mailing list
> Developer at okl4.org
> https://lists.okl4.org/mailman/listinfo/developer
> 

--